Ali Diab (@alidiabsf) 's Twitter Profile
Ali Diab

@alidiabsf

@collective CEO.

ID: 1863636709025648643

calendar_today02-12-2024 17:30:20

36 Tweet

10 Followers

11 Following

Derek Thompson (@dkthomp) 's Twitter Profile Photo

Truly the best of Twitter, in this exchange. 1. I did not know that particulate matter air pollution from cars is more about brakes, tires, and road wear than tailpipe emissions. 2. I did not know that electric cars and trucks have more emissions-efficient brake tech. 3. I

Truly the best of Twitter, in this exchange. 

1.  I did not know that particulate matter air pollution from cars is more about brakes, tires, and road wear than tailpipe emissions. 

2. I did not know that electric cars and trucks have more emissions-efficient brake tech. 

3. I
Sergey Levine (@svlevine) 's Twitter Profile Photo

I always found it puzzling how language models learn so much from next-token prediction, while video models learn so little from next frame prediction. Maybe it's because LLMs are actually brain scanners in disguise. Idle musings in my new blog post: sergeylevine.substack.com/p/language-mod…

Benjamin Todd (@ben_j_todd) 's Twitter Profile Photo

Why can AIs code for 1h but not 10h? A simple explanation: if there's a 10% chance of error per 10min step (say), the success rate is: 1h: 53% 4h: 8% 10h: 0.002% Toby Ord has tested this 'constant error rate' theory and shown it's a good fit for the data chance of

Why can AIs code for 1h but not 10h?

A simple explanation: if there's a 10% chance of error per 10min step (say), the success rate is:

1h: 53%
4h: 8%
10h: 0.002%

<a href="/tobyordoxford/">Toby Ord</a> has tested this 'constant error rate' theory and shown it's a good fit for the data

chance of
Rohan Paul (@rohanpaul_ai) 's Twitter Profile Photo

It’s a hefty 206-page research paper, and the findings are concerning. "LLM users consistently underperformed at neural, linguistic, and behavioral levels" This study finds LLM dependence weakens the writer’s own neural and linguistic fingerprints. 🤔🤔 Relying only on EEG,

It’s a hefty 206-page research paper, and the findings are concerning.

"LLM users consistently underperformed at neural, linguistic, and behavioral levels"

This study finds LLM dependence weakens the writer’s own neural and linguistic fingerprints. 🤔🤔

Relying only on EEG,
Jianren Wang (@wang_jianren) 's Twitter Profile Photo

(1/n) Since its publication in 2017, PPO has essentially become synonymous with RL. Today, we are excited to provide you with a better alternative - EPO.

Yuchen Jin (@yuchenj_uw) 's Twitter Profile Photo

Our Cornell CS intern told me that he used ChatGPT for all his assignments, then spent a week studying really hard to pass the final. I couldn’t help but think: How did I do 4 years of uni, and 5.5 years of PhD without ChatGPT? Honestly grateful I learned CS before LLMs. If I